Nonparametric approaches for drought characterization and forecasting
Publisher
The University of Arizona.Rights
Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.Abstract
Drought characterization and forecasting are extremely important in the planning and management of water resources systems. In a transboundary region, especially, sustainable water use and water rights are major issues among communities and countries during droughts. For instance, the drought of the 1990s in the Conchos River Basin in Mexico resulted in the severe reduction of its inflows into the Bravo/Grande River. Droughts in the Conchos River Basin affect many communities in the Lower Bravo/Grande River Basin, and become a controversial issue between the United States and Mexico. This research focuses on the application of nonparametric methods to characterize and forecast droughts using a drought index like the Palmer Drought Severity Index. Nonparametric methods allow more flexible approaches to hydrologic problems in practice by approximating a function of interest. Based on a nonparametric probability density function estimator, comprehensive approaches for the evaluation of drought characteristics at a site and over a region are proposed in this study. Using a kernel density estimator, a nonparametric methodology is proposed for the synthetic generation of hydrologic time series. Based on the synthetic data, a nonparametric approach is introduced for estimating the bivariate characteristics of drought. A kernel density estimator is useful in the estimation of the probability density function and the cumulative distribution function in two dimensions of drought properties. A methodology for the regional characterization of droughts is developed using the point drought properties. The nonlinear and nonparametric features of artificial neural networks provide a useful framework in forecasting hydrologic time series. A conjunction model is presented in this study in order to improve forecast accuracy for time series of droughts. The conjunction model is a hybrid neural network model combined with dyadic wavelet transforms. The overall results presented in this study indicate that the proposed methods are useful for identification of droughts, evaluation of drought characteristics, and prediction of drought occurrence. The appropriate evaluation and accurate prediction of droughts allow water resources decision makers to prepare efficient management plans and proactive mitigation programs which can reduce drought-related social, environmental, and economic impact significantly.Type
textDissertation-Reproduction (electronic)
Degree Name
Ph.D.Degree Level
doctoralDegree Program
Graduate CollegeCivil Engineering and Engineering Mechanics